Featured

Bubble & Build: The 2025 MAD (Machine Learning, AI & Data) Landscape

A market map + 25 crisp ideas for an over‑heated, heads‑down year in AI & data

Anyone who has been near AI this year has felt both: the froth of record funding and the grind of real deployment. Bubbles can be silly; they also finance railroads. 2025 is the year AI shifted from chatbots to systems that actually do work: agents with tools and memory, wired into governed data, powered by reasoning models. That tension—speculation funding infrastructure—frames everything that follows.

Welcome to the 2025 MAD (Machine Learning, AI & Data) Landscape, our eleventh edition since 2012 (an almost annual effort, prior editions here). This is the biggest redraw in recent memory. We made the editorial decision to substantially cut the logo count—from a peak of 2,000+ last year to ~1,150—to make the map legible and, frankly, possible amid an explosion of new companies and products. We also gave more space to the hyperscalers and pure-play category leaders (NVIDIA, Databricks, Snowflake, OpenAI, Anthropic etc) to reflect where market share, distribution, and developer gravity sit today.

Structure-wise, we made major edits. We deleted a few sections (notably folding the former “open source” box into the broader map as open weights/OSS now permeate every layer) and added others, for example an explicit agent stack (agent platforms, agents infra/tooling) and local AI (local/on device LLM runtimes). The result is a cleaner flow from data to infra to ML/AI to agents/applications.

As in recent years, the landscape is available both as a:

With that, here’s our roundup of the 25 ideas for 2025: what’s breaking through, what’s consolidating, and where the next wave of value is likely to accrue.

(Thank you: as every year, the MAD landscape is a team effort.  Major thanks to the FirstMark crew – Aman Kabeer (co-author), Leah Levine (logos/logistics), Ben Winn (promotion) and Ryan Sullivan (blog refresh), as well as Paolo Campos (PDF design) and Jonathan Grana (interactive version, which he does as a side/promotional project at Go Fractional)

Macro & Markets

1/ Bubble dynamics without the brakes. The market is frothy again, but this isn’t 1999.   Money is plentiful and valuations often stratospheric, with a clear “AI premium”, especially around agents, frontier AI, and anything growing fast (retention concerns notwithstanding). As tends to be the case in paradigm shifts, capex/opex are front-loaded. Demand will need to materialize in a big way if the space is to stick the landing, but habits often take time to change and adoption is uneven. Add the human reality: many teams are running at 996-ish intensity—frenetic sprints that accelerate shipping velocity but raise burnout risk. On the other hand, the paradigm shift is clear, revenue is real, growth is often impressive, unit economics are legible, and the denominator is bigger.  The paradox of 2025 is that hype and fundamentals are both up; history suggests fallout can arrive before the payoffs. The year’s tone is acceleration, with nerves attached.

2/ Fragility: circularity and customer concentration. Under the big growth numbers, a lot of money is looping around a small set of players.  Some deals look circular: OpenAI inks massive GPU buys with NVIDIA while NVIDIA commits giant investment back into OpenAI; plus a multibillion chip pact with AMD that includes an option for OpenAI to buy a stake. Similar patterns extend across the stack: industry financing and supply agreements increasingly tie model labs, chipmakers, clouds and AI startups into webs of mutual dependence, prompting “roundtripping” concerns. Customer concentration makes the AI ecosystem less shock-resilient: a large share of spend runs through a short list of hyperscalers and frontier labs, while several breakout vendors lean on a handful of outsized customers – great until a growth pause or policy change ripples through. API terms, safety updates, or rate limits can impact numbers overnight. The engine is running hot, and the load is unevenly distributed.

3/ Big picture fuzzy, near-term very real: Credible voices are split on whether progress is plateauing or we’re simply missing the next exponential; paths to AGI/ASI remain undefined and the very definitions are fuzzy; even the doomer drumbeat has quieted. Meanwhile the short-to-medium term feels unmistakably concrete: an avalanche of AI slop across video, text, and code is arriving alongside ever more pressing concerns about jobs: how much changes, how fast, and for whom. The stakes are immediate even if the endgame isn’t. And, as usual, human, political, and societal responses are lagging the speed of the technology.

4/ Labs vs. incumbents: different balance sheets, same race. The fight to dominate AI is fiercer than ever, and the field is uneven. Big Tech has massive distribution, huge product suites, and budgets to bundle, wait out cycles, and grind. Google clearly regained momentum in 2025 with a string of headline AI launches, and Meta ramped ambitions with its SuperIntelligence Lab; both run on hugely profitable cores and near-infinite balance sheets. Independent frontier labs, by contrast, need step-change breakthroughs to justify valuations. New names—SSI, Thinking Machines, Reflection—joined the top tier, right as agent/reasoning heat rose (and distribution hurdles remained). OpenAI remains the clear leader and keeps raising war chests; Anthropic isn’t far behind, but how long can capital run at those levels? Users win either way: incumbents bundle, labs dazzle—while M&A stays the bridge between missing pieces and market access.

5/ IPOs & public comps: the window is open (selectively).  CoreWeave’s March debut did what the market needed: a clean AI-infra IPO that’s traded well since. Palantir is the lightning-rod comp, riding a premium EV/NTM multiple (~80–90× lately), which should embolden late-stage filers. Next: Cohere says it could IPO “soon,” Dataiku has picked banks, while Cerebras withdrew its S-1 after a fresh raise.  Meanwhile, the top 10 or so private AI players have little incentive to go public, given access to capital and strategic flexibility; but if/when Databricks (> $100B private) and the frontier labs (OpenAI, Anthropic) do eventually go out, expect bonanza, record-breaking IPOs.

6/ M&A: consolidation and talent wars. Big players tried to build full agent stacks, found it harder than it looked, then went shopping—yet even headline deals stalled (Adobe–Synthesia, SoftBank–Agility), so “buy vs. build” isn’t automatic. The wins were surgical: ServiceNow–Moveworks ($2.85B) for enterprise agents; Salesforce–Informatica (~$8B) to firm up the data control plane. Data infra is merging from within: dbt Labs and Fivetran (all-stock; ≈$600M ARR) bring ingestion and transformation under one roof. The loudest story, though, is talent and acqui-hires. Meta, in particular, went on the offense: it took ~49% of Scale AI for ~$14–15B to bring in Alexandr Wang on its superintelligence push, then set nine-figure comp markers and poached OpenAI researchers, triggering a retain-at-all-costs spiral. Bottom line: 2025 is the year of precision tuck-ins, team buys, and creative structures; true megadeals remain rare under integration risk and antitrust glare.

Research & Frontier

7/ Reasoning + RL is the frontier. The biggest leap this year wasn’t a bigger transformer; it was training models to spend compute on thinking. Reinforcement learning for reasoning—popularized by DeepSeek R1 and “o-series” style models that allocate tokens to deliberation moved the needle across math, code, and multi-step planning. Curriculum design, reward design, and tool-use feedback loops matter more than raw model size. RL is not a silver bullet —bad rewards still teach bad habits—but scaled correctly, it brings tremendous punch to pre-training. The next challenge is generalization beyond code and math to messy real-world work where “right” and “wrong” aren’t always crisp; here, richer signals matter, from business outcomes to human feedback and new benchmarks like GDPVal, which score end-to-end task chains. 

8/ Is AI slowing down? Dissent keeps us honest. Some top researchers—including guests on our MAD Podcast (Sholto Douglas, Julian Schrittwieser, Jerry Tworek)—say there’s still plenty of low-hanging fruit and years of progress ahead using the current pre-training + RL paradigm. Others urge caution: Andrej Karpathy says “agents are a decade away”; Rich Sutton’s Bitter Lesson argues that general methods plus compute beat hand-tuning; Yann LeCun pushes world models and self-supervised prediction as a different path. The debate is healthy: less leaderboard theater, more ablations, red-teaming, and real-world tasks. 

9/ Fast-moving frontiers: AI doing inventive science; robotics. We’re seeing “Move 37” ideas in the lab—models proposing non-intuitive hypotheses and paths humans wouldn’t try first. AlphaFold 3 moved into biomolecular interactions; GNoME surfaced ~2.2M plausible crystals; and Yale × Google’s Cell2Sentence-Scale 27B flagged a potential cancer-therapy pathway from single-cell data. Beyond bio, robotics is accelerating: robotics foundation models (vision-language-action policies trained on large, pooled datasets) are improving transfer across robots and tasks, while mobile manipulators log more real-world hours and autonomous lab rigs tighten the design–build–test loop. Could AI deliver Nobel-level breakthroughs, or field robots that reliably do useful work? Both feel closer each quarter.

10/ Open source (open weights) endures—through a bumpy year. DeepSeek’s R1 moment (and open-weights derivatives) set the tone, but Llama 4 underwhelmed and Meta signaled a tighter stance on permissive releases. Mistral had swings, then regained momentum; Qwen3 quietly became the “good-enough” workhorse in many stacks. On the upside, AI2 kept shipping real assets (OLMo/OLMo-2, Dolma-class data), and Reflection AI’s funding revived the “U.S. DeepSeek” narrative. Enterprises still want control and residency; startups still want margin. The future seems hybrid: route to open source when you can, spike to frontier when you must. With NPUs landing everywhere, small models will play an important role; the healthiest stacks stay plural—open and closed, cloud and device, large and small —without religious wars or vendor lock-in.

Geopolitics

11/ China assembles a parallel AI stack. China is building an end-to-end path that leans less on NVIDIA and CUDA: Huawei Ascend 910B/910C under a growing software layer (CANN, MindSpore), topped by homegrown models (DeepSeek, Qwen3, Kimi, ERNIE, etc) tuned for local data and policy. Since the DeepSeek moment, it’s been a big year for Chinese models, with Qwen and Kimi expanding in production, not just “good enough,” but competitive in several domains. Export controls slowed but didn’t stop progress; localization became a feature, signaling technical decoupling: compatible, increasingly self-reliant, and in places front-rank. 

12/ Sovereign AI goes from slogan to procurement. “Build local models on local compute” now has hardware, budgets, and real buyers behind it. The U.K. switched on Isambard-AI and finished its grid hookup; IndiaAI crossed 34,000+ GPUs and started subsidized allocations; Gulf states keep scaling national “AI factories” via G42 × Cerebras (Condor Galaxy). Europe is nurturing champions—Mistral now with ASML in its corner—while OpenAI rolled out EU/U.K. data residency to meet sovereignty asks.

13/ Energy becomes the new compute chokepoint, and nations notice. Power, not GPUs, is the new bottleneck. Datacenter location decisions now follow megawatt contracts, water rights and grid interconnects. Government court AI factories like they court fabs. Expect sovereign PPAs and nuclear/renewables co-location (Isambard-AI grid hookups in the UK; Google – TVA/Kairos SMR iplitos; Microsoft-Helion Fusion PPA). Power-first incentives will shape where models are trained and which regions win the AI buildout. Export controls still matter, but kilowatts now set the timelines.

The Business of AI

14/ Distribution beats invention (again). A whole generation of AI-native  startups is growing faster than we’ve ever seen. Products go viral on social, boards continue to fret about AI and curiosity fuels a wave of trials and tinkering. The open question is durability: true ARR or experimental revenue that churns? Incumbents often hold the distribution edge: assistants bundled with iOS/Android, Windows Copilot, Chrome, Salesforce Einstein, ServiceNow Now Assist—but not always. Partnerships and integrations can bend the curve: Cursor deepens into VS Code; Supabase rides the Lovable wave; quieter winners seep into IDEs, CRMs, and docs. Products find success by being present at creation moments (writing, coding, filing a case) where embedded beats merely “better,” and expansion follows real usage.

15/ Margins & pricing: land-grab to landing the plane. When usage climbs and customers want the newest, smartest models, costs jump fast. Hard truth: if you sit on other people’s frontier models, growth can flip you into negative gross margins—the Windsurf to Cognition story is a warning. VC money can fund the land-grab, but it won’t cover bad unit economics forever. AI startups are adapting: default to smaller, cheaper models, reserve capacity for peaks, and cache aggressively. The dominant approach is becoming price to outcomes—per case closed, per ticket resolved—with options for guaranteed throughput, so revenue tracks real results, not chatter. Winners pair cost discipline with pricing that meters actual value.

16/ Enterprise AI: deployment lags demos (but it’s landing). Enterprise deployments move slower than cool demos on social media.  Buyers want agent governance, citations, provenance, PII handling, audit trails, and tight ties to enterprise systems before switching on any level of autonomy. There is real progress on defining and implementing use cases: AI customer service, AI coding, and internal chatbots are the obvious wins; many industry- or company-specific plays will need customization, data plumbing and policy work to fully emerge. But we’re past the “Accenture phase”, and the arc is now visible—copilots to narrow agents to managed automations – and demand is firming up. Into that demand, incumbents have a distribution advantage, shipping “agent platforms” inside CRM/ERP/ITSM (Salesforce, ServiceNow, Microsoft), bundling guardrails, telemetry, and approvals in one place. But as always, never underestimate startups.  Overall, the Global 2000 enterprise market is warming up to buying and deploying AI in earnest, just not boiling yet.

AI Infrastructure 

17/ NVIDIA dominates, but diversification is real. Blackwell GB200 racks remain the reference point, yet buyers are now adding Google TPUs, AMD MI350 and, in specific footprints, Intel Gaudi 3. With rack-scale design driving TCO, many shops mix vendors for price/perf and supply, and run heterogeneous clusters under smarter schedulers—not a single-vendor monoculture. 

18/ Local AI rises: device, near-edge, and private clouds. New NPUs in laptops and phones push real work onto the device: fast, multimodal, and private by default. When tasks are too big, they spill to nearby or vendor-run “private clouds” (e.g., Apple’s Private Cloud Compute) instead of generic public endpoints. Tools like LM Studio and Ollama make local models click-to-run. On-device handles snappy UX and personal context; cloud handles heavier reasoning and shared memory. In factories, clinics, and cars, near-edge boxes protect bandwidth, privacy, and uptime. The best products hand off smoothly across device, edge, and cloud.

19/ The agentic stack becomes an infrastructure layer. Beneath apps sits a new runtime: planners and tool-calling, structured outputs and function catalogs, long-/short-term memory (vectors, graphs), sandboxed tool execution, approvals, and stateful orchestration. Around it: eval harnesses for tasks, policy/guardrails, traces and cost telemetry, dataset/version control, and rollback. What looked like “app glue” in 2024 now resembles a platform tier with its own SLAs and procurement line.

20/ Compliance, security, and red teams are foundational. Security and compliance are not a checkbox, it’s the price of running AI in production. Updated guidance (e.g., OWASP’s LLM Top 10, prompt-injection playbooks) set the bar: show where data came from, log prompts/tools/decisions, enforce policy, and prove you resist jailbreaks. Enterprises expect attestations, audit trails, and clear “break-glass” procedures, wired into the same layer as serving and storage. If it can’t be evaluated, traced, and governed, it isn’t infrastructure.

Data Infrastructure 

21/ End of an era, start of a merge. The “modern data stack” unbundling is giving way to consolidation: dbt Labs and Fivetran are combining, while platforms like Databricks keep covering the waterfront (batch and streaming, vector and graph, feature stores, governance) by equal parts build and buy. The frame shifts from “warehouse vs. lakehouse” to object storage plus open tables and a neutral catalog as the control plane. Modeling, movement, features, eval datasets, lineage, and policy are fusing with AI serving and the agentic runtime. In effect, data infrastructure and AI infrastructure are collapsing into one plane; the seams are where value leaks.

22/ Yet the data fundamentals remain more important than ever. Robust tables and catalogs, quality and lineage, and low-latency query engines have become prerequisites for agents, retrieval, and eval-first CI—not afterthoughts. Graph- and vector-augmented retrieval is moving from blog post to pattern, observability now spans prompts, tools, and cost, and compliance sits alongside performance in the same plane. The space has fresh vigor: ClickHouse’s rise in real-time analytics (now with vectors) signals demand for speed at scale, while local and edge stacks still need clean contracts back to cloud memory. Data isn’t fading; it’s been promoted to AI’s control surface.

Applications & Agents

23/ Big Labs and Platforms move up the stack. The frontier labs and incumbents aren’t content with being just model APIs: OpenAI, Anthropic, and Google/Gemini keep shipping app-layer products: voice assistants, desktop apps, team plans, and workflow builders stitched to mail, docs, and CRM. That creates platform risk and frontal competition: when the model vendor owns the surface and the bundle, it can ship into your lane tomorrow. OpenAI pushed farthest, recruiting domain experts (e.g., ex-bankers) to teach workflows, adding commerce rails inside ChatGPT, and launching a ChatGPT-first browser—while Anthropic deepened team/project flows and shipped Claude Code; Gemini tightened its consumer and Workspace surfaces. Meanwhile, models have absorbed big chunks of the “wrapper” layer: first-party structured outputs, function calling, memory, browse/code/vision/voice tools, lightweight automation, even commerce. Users get speed—capabilities landing where they already work. For startups, the wrapper cycle went thin → thick → thinner again: early UIs grew into real products (data bridges, workflow, compliance) only for the platforms to pull many features into the core. The lane that remains is narrower but real: deeply specialized workflows tied to systems of record, proprietary data/logic, and surfaces the platforms don’t, or won’t, prioritize. And for clarity: Microsoft has long lived at the app layer; the new encroachment story is led by OpenAI, with Anthropic and Gemini close behind.

24/ Vibe coding becomes the hit of 2025. Coding agents jumped from novelty to daily habit—reading repos, spinning sandboxes, planning changes, opening PRs, running tests, narrating diffs—and even “video coding” demos now show agents manipulating UIs from screencasts. Adoption has been breathtaking: Cursor and Claude Code are widely cited as among the fastest-growing dev tools ever, with reported nine-figure ARR trajectories within months. The craft shifted from autocomplete to directing and reviewing, and the stack broadened (GitHub Copilot, Sourcegraph Cody, Codeium/Windsurf, Devin, etc) toward end-to-end workflows. On the product side, Vercel v0, Lovable, and Replit turned “describe, then ship” into a production loop for tiny teams. The open question is stickiness, especially for non-professional developers, yet early cohort behavior suggests these habits may be as durable as search for coding.

25/ Modalities light up. Image, video, and voice hit a new gear: Veo3, Runway and Sora drove cinematic generation; ElevenLabs and Synthesia made high-quality voice and avatar work routine; real-time voice agents hold fluid conversations and drive tools. Vision models now parse UIs, charts, and field photos without brittle templates, and video editors jump from clips to storyboarded scenes with provenance. Meanwhile, world models—from Genie 3 to new work out of Fei-Fei Li’s group—aim to perceive and act in interactive environments, blurring creative and operational software. The bar moved from “can it caption?” to “can it perceive, plan, and act across modes, reliably?” 2026 will be a big year for modalities.

Closing thought

The 2025 MAD Landscape is a map of a market doing two things at once: bubbling and building. We redrew it to reflect reality—fewer logos, more gravity—where hyperscalers and pure-play leaders anchor the edges, agents and the data/control plane meet in the middle, and energy, not just GPUs, sets the tempo. The story lines rhyme across the map: labs climb into apps as open weights stay resilient; data and AI infra merge; enterprise deployment lags demos but is landing; coding agents become a daily habit. From here, the horizon is bigger than any single release: if we align distribution, margins, governance, and kilowatts, intelligence becomes infrastructure—and the next wave turns into compounding progress that lifts whole industries.

Featured

Quick S-1 Teardown: CoreWeave

In the tradition of previous quick S-1 teardowns (Snowflake, Palantir, Confluent, Klaviyo, Cerebras, etc), some quick notes on the CoreWeave S-1 from my colleague Aman Kabeer and I.  As in prior efforts, this is not meant to be 100% comprehensive (and it’s certainly not investment advice!).

The CoreWeave IPO is going to be fascinating to watch: partly because it is undeniably exciting, and partly because it is not going your standard tech IPO.  It  presents a profile that in some ways is typical of a hyper-growth tech unicorn (explosive growth, large losses, dual-class stock structure with Class A/B shares​, etc.), but in other ways, it is very unlike most tech IPOs of the past. Its specialized business model, heavy infrastructure focus, heavy customer concentration (Microsoft), reliance on big partners (NVIDIA), financial structure ($7.9B in debt) and unusual risk factors make it a unique case that blends the characteristics of a cloud provider, a hardware company, and a startup riding an exciting but also sudden and unproven market wave. 

SOME KEY TAKEAWAYS

Frivolity: Crypto pivot, NJ in da house: For all the jokes on social media about founders, startups and VCs pivoting from Web3 to AI, CoreWeave is an example of a business that started as a crypto mining operation, stockpiled GPUs, and pivoted to AI with spectacular success.  

In the same (frivolous) vein, for all the “all AI is in SF” mantra, CoreWeave is headquartered in… New Jersey.

The First Generative AI IPO:  Depending on pricing, there’s likely going to be tremendous interest in the IPO.  This is in part because of the well-documented dearth of tech IPOs, but most importantly  because it’s the first IPO of the Generative AI era (Cerebras, as far as we know, is still stuck in CFIUS review of their relationship with G42).  

There are very few “AI pure plays” on public markets – Palantir is one example (cue in the never ending discussion as to whether they are truly an AI company), as is (even more arguably) C3 AI.  Other than that, the way to “play” Generative AI has been to invest in Mag 7 companies.

It is no accident that the first companies to file S-1s in the Generative AI era are infrastructure companies.  The market has been forming supply-first  (chips, data centers, foundation models), with the major hope that the demand side surfaces equally meaningfully in years to come. 

Not a Real Estate Play: A negative take on CoreWeave and comparable companies one would often hear in tech circles is that the company is a “real estate play”, with limited technology and software.  The argument seemed to be supported by the fact that the co-founders of the company come from a financial, rather than technological, background.

Continue reading “Quick S-1 Teardown: CoreWeave”

New Investment: Gradium

A major new entrant in voice AI: Gradium.

If we were designing computers from scratch today, the default interface probably wouldn’t be a keyboard. It would be voice.

Yet most AI products still look like a text box. We type, we click, while the microphones in our phones and laptops mostly sit idle. Voice is the most natural interface we have, and probably the most underserved.

For all the progress in AI, voice suffers from unatural latency. It struggles with taking turns. Accents occasionally sound funny. And, most importantly, voice is often prohibitively expensive to be deployed at scale.

A big reason is that “voice AI” is not one problem, but a stack of narrow, hard problems: ultra-low-latency streaming, high-fidelity synthesis and prosody, robust speech recognition, instant translation, voice conversion and cloning, safety, and more. Each layer is its own specialty. There are only so many people in the world who have actually built these systems end-to-end at scale.

Enter Gradium.

Gradium is building audio language models – an audio-native counterpart to LLMs – to power real-time voice applications across use cases: gaming, AI agents, education, healthcare, customer experience, and beyond. The goal is to become the technical backbone of global voice technology, breaking the old trade-off between quality, latency, and cost so that truly natural voice becomes ubiquitous, not a premium feature.

We’re excited to announce that FirstMark is co-leading Gradium’s $70M seed round with Eurazeo alongside DST Global, Eric Schmidt, Xavier Niel, Rodolphe Saadé, Korelya, Amplify Partners and others.

Gradium was founded in September 2025 by Neil Zeghidour (Meta/Google DeepMind), Olivier Teboul (Google Brain), Laurent Mazare, Google DeepMind/Jane Street) and Alexandre Défossez (Meta). Collectively they invented and open-sourced neural audio codecs and audio language models, and used this technology to power the very first voice cloning, text-to-music generation and speech-to-speech translation. They then created Kyutai, a non-profit lab pushing the frontiers of multimodal LLMs, in particular releasing the first real-time conversational model in 2024.

This is one of the highest-density clusters of deep generative audio expertise, anywhere. They’ve already moved at unusual speed for a foundation model company, generating revenue within weeks of founding.

Gradium is based in Paris, with a presence in San Francisco. We continue to be very excited about European AI startups – in the vein of our investments in Synthesia, Dataiku and Pigment, which have grown into global category leaders – and Gradium fits squarely in that story. We’re thrilled to partner with Neil and the team from day one, alongside our friends at Eurazeo and a great group of co-investors.

If you’re building anything voice-native – agents, games, learning tools, healthcare or something entirely new – you can start exploring the platform at gradium.ai.

Announcing the 2025 MAD Landscape (Coming Soon!)

It’s almost here: the 2025 MAD (Machine Learning, AI & Data) Landscape.
This year’s edition highlights the companies, trends, and open-source movements shaping AI—from foundation models and agents to data infrastructure, security, governance, and real enterprise adoption.

We’re putting the finishing touches on the map and analysis now. If you have comments, corrections, or suggestions—including companies or projects we should consider—please email mad@firstmark.com.

Stay tuned for the full release and deep-dive write-up—coming soon!

AI in 2024: Never a Dull Moment

In AI, it’s been the year of big, bigger and biggest. My colleague Aman Kabeer and I just did a fun episode of The MAD Podcast where we discussed what we’re seeing in the market, our favorite trends and new stories, and where we see things going.

Here’s the video, and I summarized below a few highlights:

Continue reading “AI in 2024: Never a Dull Moment”

Quick S-1 Teardown: Cerebras

Look up in the sky! It’s a bird! It’s a plane! It’s… an IPO. The Cerebras S-1 filing is interesting in many ways, but certainly one is that, well, it is an (upcoming) IPO in the first place.  In a context where tech IPOs have been at an all time low, with a very modest uptick in 2024 (Reddit, Rubrik, etc), the fact that  a VC-backed tech startup has filed is rare enough to be exciting and newsworthy on its own. 

The other unmistakable part of the filing is that Cerebras is a “pure play AI” company, in a context where there’s been a dearth of such companies in public markets, outside of Palantir and arguably a couple of others, like C3 AI or recent entrants like Tempus AI and Astera Labs. For the most part, public market investors have had very limited options to play the Generative AI wave: essentially NVIDIA, and indirect bets on AI through the hyperscalers. (This scarcity of AI stocks and to some extent, data infra stocks, is a reality we captured in 2021 through our MAD Public Company Index, that will soon be worth updating as hopefully more IPOs happen).

Continue reading “Quick S-1 Teardown: Cerebras”

Is SaaS dead?

Many SaaS stocks have been getting clobbered in public markets. Some see the “end of software“.

Is SaaS dead?

What seems to be happening:

  • tough macro, cost cutting
  • AI sucking the air out of the room
  • SaaS vendors perceived as “last generation” despite best efforts to add AI quickly
  • enterprise budgets for AI are not net new, they’re taken from somewhere (SaaS budgets cut)
  • Bulk of budgets going to OpenAI/Azure etc because low hanging fruit to “do AI” (knowledge bot, coding)
  • for the more specialized enterprise apps, customers feel like they can/should “build” internally rather than “buy”

What happens next:

  • customers realize that “build” is a headache, not always the best option
  • OpenAI / Azure etc can’t / doesn’t want to build hundreds of problem specific/ vertical specific apps
  • Takes time, but legacy and new SaaS companies truly become AI-first (not just marketing), abstract away complexity of deploying LLMs
  • macro environment eventually rebounds
  • AIaaS becomes the new SaaS – what is old is new (unedited Sat morning thoughts)
  • Question is what happens to all current SaaS unicorns and public companies as this transition happens

(unedited Saturday morning thoughts)

MAD 2024: Trends in AI & Data (video)

As a companion to the 2024 MAD (ML, AI & Data) Landscape (blog post, PDF, interactive website), my colleague Aman and I had a fun chat about some key trends we see in data and AI.

Some topics we covered:

  • The impact of open source in AI
  • The future of AI agents
  • Where are we in the AI hype cycle?
  • The emerging AI stack
  • Will AI kills SaaS?
  • Is the Modern Data Stack dead?
Continue reading “MAD 2024: Trends in AI & Data (video)”

Full Steam Ahead: The 2024 MAD (Machine Learning, AI & Data) Landscape

This is our tenth annual landscape and “state of the union” of the data, analytics, machine learning and AI ecosystem.

In 10+ years covering the space, things have never been as exciting and promising as they are today.  All trends and subtrends we described over the years are coalescing: data has been digitized, in massive amounts; it can be stored, processed and analyzed fast and cheaply with modern tools; and most importantly, it can be fed to ever-more performing ML/AI models which can make sense of it, recognize patterns, make predictions based on it, and now generate text, code, images, sounds and videos.  

The MAD (ML, AI & Data) ecosystem has gone from niche and technical, to mainstream.  The paradigm shift seems to be accelerating with implications that go far beyond technical or even business matters, and impact society, geopolitics and perhaps the human condition. 

There are still many chapters to write in the multi-decade megatrend, however.  As every year, this post is an attempt at making sense of where we are currently, across products, companies and industry trends. 

Here are the prior versions: 2012, 2014, 2016, 2017, 2018, 2019 (Part I and Part II), 2020, 2021 and 2023 (Part I, Part II, Part III, Part IV).

Our team this year was Aman Kabeer and Katie Mills (FirstMark), Jonathan Grana (Go Fractional) and Paolo Campos, major thanks to all.  And a big thank you as well to CB Insights for providing the card data appearing in the interactive version. 

This annual state of the union post is organized in three parts:

  • Part: I: The landscape (PDF, Interactive version)
  • Part II: 24 themes we’re thinking about in 2024
  • Part III: Financings, M&A and IPOs 
Continue reading “Full Steam Ahead: The 2024 MAD (Machine Learning, AI & Data) Landscape”

New Investment: Lago

Billing infrastructure has been a vexing issue for generations of software and Internet companies.  It is mission critical infrastructure and as such, feels like a problem that should have been sold a long time ago. Yet ask any company and you’ll generally get the same reply: they’re dissatisfied with their billing system, which doesn’t offer the level of flexibility to address their specific needs and the many edge cases that inevitably pop up.

The problem is only getting worse as the software industry transitions from subscription-based to consumption-based revenue models. What started as a trickle is becoming mainstream, as usage based pricing unquestionably builds better alignment with customers. This transition will only be accelerating, as an entire generation of new AI companies coming online that are almost all using that pricing model.

Today, we’re excited to announced our Series A investment in Lago, the leading open source metering and usage-based billing company. Lago offers both a self-hosted and cloud, scalable and modular architecture, and found strong product market fit both as an open source project and a commercial product.

Our investment thesis is pretty simple:

  • Usage-based native: Lago is natively focused on the specific problem of usage-based billing and as such, incredibly well positioned to be the default solution for a whole new generation of companies, including in particular AI companies – it is not accident that customers already include AI unicorns like Mistral and Together AI
  • Open source: other than the fact that we love open source infra in general (as evidenced by our investments in Cockroach Labs, ClickHouse, Astronomer, SurrealDB, Quickwit, etc), open source is a particularly formidable advantage for the billing space in general, as it enables a level of extensibility and composability that is uniquely suited to edge cases. Lago is a natively open source company with strong OSS traction and a vibrant community
  • Europe/US: Lago is one of those bi-continental companies we love at FirstMark, with product and tech based in the vibrant Paris tech ecosystem and GTM, over time, in the US. Particularly as its ambitions go beyond “just” billing, Lago will be building a tremendous amount of product over the next few years, and will be well served by the relative cost advantage that a European location offers.
  • Team: Needless to say, first and foremost we love the Lago team, led by co-founders Anh-Tho and Raffi, both incredibly thoughtful and gritty. The Lago team comes from a place of deep industry knowledge, having built the entire billing infrastructure at fintech unicorn Qonto.
  • Bonus: getting to hear my colleague Aman repeat “open source metering and usage-based billing”, an area he’s particularly passionate about, thousands of times with equal enthusiasm

We very much look forward to working with Lago, and we’re excited to join a great group of prior investors including our friends at New Wave, SignalFire and some of the “French AI mafia” like Clement Delangue from Hugging Face and Romain Huet from OpenAI.

Of course, Lago is hiring.

My VC resolutions for 2024

1) Be an AI leader: Buy at least one share of Open AI in that employee secondary at an $86B valuation – then change website and all social profiles to “early believer and investor in Open AI”

2) Leverage the network: In conversation, drop frequent references to “Sam”, “Satya” and “the Besties” – I put in the hours building the relationship, by liking their tweets and listening religiously to the All-in podcast, now people need to know I’m tight with those guys

3) Add value: Formula 1 is a major sport in the US now, and founders in my portfolio will want know that I “get it”. Plan on attending several Grand Prix in 2024. While in Miami, Vegas or Monaco, send founders energizing texts like “Do you have the DRIVE TO SURVIVE?”, or “What would it take for us to be in POLE POSITION next year?”. They may not reply, but I know they will appreciate.

4) Refine investment thesis: Ok, so, I haven’t really done a new deal in over a year. What do other VCs invest in these days? AI is cool but exhausting, changes like every day. Defense tech seems hot, and blowing sh*t up is fun, so maybe? Heard about “nuclear fusion” and “superconductor” – ask ChatGPT to explain those “like I’m 5”, then tweet that out, to establish thought leadership

5) Inspire: Founders love it when VCs tweet during weekends and holidays things like “How bad do you want it?” or “Hustlers never rest!”. Pre-schedule a bunch of those tweets to automatically publish throughout the year.

6) Be a VC leader: Founders calling me all the time gets annoying, but I always have time for journalists. On the record, comment on VC firms doing layoffs or shutting down and say it’s “healthy for the industry”, but doesn’t affect me because I’m “top quartile”. Off the record, give the journalist a list of GPs who beat me on deals as examples of folks who are “in serious trouble”.

7) Stay fit: First it was kitesurfing, then it was pickleball. These days it’s jiu-jitsu a la Zuck, Elon and Lex Fridman. Possibly also padel? Gosh, few understand the level of pressure VCs are under to perform.

8) Be a master planner: Didn’t do a good job planning last year. Designer clothes showed up late at Art Basel, didn’t get my usual suite upgrade at the Crosby Hotel in SoHo, and missed the best DJ set at Slush. Do better in 2024. Upgrade my boat setup for Mykonos this Summer, and be the ultimate “man in the marina”.

Happy new year! LFG 2024!

P.S. Holding myself to the highest standard year after year, see 2023 VC resolutions, 2022 VC resolutions

New Investment: Sparta Commodities

If you scroll through the list of biggest companies in the world by market capitalization, it will come as no surprise that, in the now tech-dominated ranking, commodities companies continue to be heavily represented, with many familiar names such as Saudi Aramco, Exxon Mobil, Chevron, Shell, Petrochina or Total.

Those companies are not just large, they are also cash-generating machines: 11 of the 20 most profitable companies in the world are commodities companies.

Of course, the world of commodities (energy resources, metals, agricultural goods, etc) also manifests through commodities trading – the exchange of different assets, typically futures contracts, where investors make bets on the expected future value of a given commodity, whether for economic or speculative reasons. Alongside equities and fixed income, commodities is one of the key asset classes in financial markets.

Continue reading “New Investment: Sparta Commodities”

Quick S-1 Teardown: Klaviyo

Is this it? Are we back? Everyone in the startup and venture world has been waiting for months for the re-opening of the IPO window. After a record breaking 2021 (1035 IPOs, beating the previous record of 480 in 2020), 2022 saw a dramatic decline (181 IPOs) and 2023 so far has not been much better.

Common wisdom in the market over the last few months has been that Q4 2023 would be the time the IPO window would cautiously re-open for technology companies (recent non-tech IPOs like restaurant chain Cava being considered non-representative). And it would be crucial that some of the very best companies (the usual suspects being Stripe, Databricks and Instacart) would go out first, to pave the way for a bigger wave of quality companies right behind them.

Well, this week has been an exciting one – on Monday, ARM filed its F-1 (here) and just today (Friday August 25), both Instacart (here) and Klaviyo (here) filed their S-1s. It’s going to be exciting to see what happens this Fall in IPO land.

New IPO filings also mean fresh opportunities for the time-honored VC tradition of S-1 breakdowns, even though timing is unfortunate given summer vacation schedule – here and here.

Consistent with my general investing focus on data and ML/AI, I’m going to pick Klaviyo for this first breakdown of 2023, as it’s a heavily data-driven business. As I did in the past (see the S-1 quick teardowns for Snowflake, Palantir, Confluent, C3, nCino), this is meant as a QUICK breakdown – mostly unedited notes and off-the-cuff thoughts, in bullet point format.

Let’s dig in.

Continue reading “Quick S-1 Teardown: Klaviyo”

In Conversation with Florian Douetteau, Co-Founder & CEO, Dataiku (podcast + video)

An overnight success 10 years in the making, Dataiku, the leading enterprise AI platform targeting Global 2000 companies, was named “Partner of the Year” for data science, machine learning and AI by BOTH Snowflake and Databricks at their respective annual summit a few days ago.

The company, in which we’ve been proud investors since leading the Series A in 2016, has scaled impressively over the years, reaching $200M in ARR at the end of 2022, with a team of over 1,200 people.

It was a pleasure welcoming back CEO Florian Douetteau, for a conversation where we covered:

* Dataiku’s centralized approach to enterprise AI

* Emerging use cases for Generative AI in the enterprise

* Some leadership lessons learned along the way

Here are the links to the podcast (subscribe! give us 5 stars! etc), and the YouTube video:

Continue reading “In Conversation with Florian Douetteau, Co-Founder & CEO, Dataiku (podcast + video)”

This Week in AI: Databricks’ Acquisition of MosaicML

(This post is part of my “This Week in AI” series, which is general off-the-cuff market commentary. I’m not an investor in either MosaicML or Databricks)

$21M per employee. That’s the price Databricks is paying for MosaicML — a total of $1.3B for 62 employees (in Databricks stock, and also includes employee retention packages).

One thing is clear – if you’re going to be aggressively acquiring Generative AI startups, you’re going to have to pay up

But it may turn out to be cheap in the long term given the size of the opportunity.

That’s because, beyond any Generative AI capabilities, Databricks’ move needs to be understood in the broader context of its fierce rivalry with Snowflake.

Continue reading “This Week in AI: Databricks’ Acquisition of MosaicML”